Cyber Cube and Munich Re have published findings from a joint study into systemic cyber risks, highlighting widespread malware and cloud service outages as credible large-scale scenarios that could severely impact global business operations.

The study draws on insights from 93 cyber security professionals, offering a detailed picture of how systemic cyber events might unfold, and which mitigation measures prove the most effective.

Respondents suggested that a severe malware event could infect up to 25% of systems globally, though only 15% would likely be fully compromised. Patch management, network segmentation and data back-ups were deemed the most effective mitigations, reducing exposure and financial impacts by as much as 80% when properly implemented.

Cloud outages were another key concern, with cyber professionals expecting such events to last hours or days. A one-day outage of a critical cloud service provider could result in financial losses equal to 1% of annual revenue, with wide variation depending on industry, company size and contingency planning capabilities. Multi-region architecture was seen as the most effective defence, while switching between providers during an outage was considered unfeasible.

The report also flags IoT devices as the most pressing new concern, with large language models already influencing the threat landscape, and artificial general intelligence considered a longer-term risk.

Unlike today’s AI, which is highly specialised and limited to predefined tasks, the theory is that AGI would have broad cognitive capabilities – much closer to human intelligence.

“By sharing the findings of our study on systemic cyber risks, we aim to provide a more nuanced view of how systemic cyber events might unfold and the factors that drive wide variation in risk exposure across firms,” said Jon Laux, vice-president of analytics at Cyber Cube.

The results of the study have been used to improve Version 6 of Cyber Cube’s Portfolio Manager, and informed enhancements to Munich Re’s internal models.

---